Daan B wrote:
>Good luck finding a well taken care for used 1Ds3... I won't sell you mine though
Oh go on, pretty please ;-)
BTW Do you need clean low ISO shadows for that kind of groupshots?
It's always a benefit, but these shots are to show why it helps to have surplus resolution, as well as surplus dynamic range. Ideally I would only use one kind of camera, for sanity. I don't really want one camera for DR and one for rez.
Daan B wrote:
Brainiac... Don't tell me the only reason you sold the 1Ds3 is because of it's weight and size...
Pretty much. If you have a camera around your neck while walking for 14 hours (e.g. photographing a wedding) that matters. If you prefer to shoot lightweight primes, then you want 2 or 3 cameras around your neck for 14 hours. Size and weight matter a lot to me. It's an ongoing problem: how to get the best images while carrying the least weight. 2 or 3 5D2's I can live with, whereas 2 or 3 1Dx's will be the death of me. I also shoot a lot of shots one handed without using a viewfinder. The 1Dx's produce quite serious wrist and finger strain. I once saw an old lady wedding photographer who pulled her gear around behind her in a wheeled basket. She was also using a Contax N1. What a pro.
brainiac wrote:
Quite right - high iso is more important to me, and image quality at high iso is generally much lower than low iso, so that is where customers notice problems in quality.
BUT, often enough I have to photograph a bride in a white dress in sunlight and her face in shadow. A significant number of my pictures suffer from blown out highlights because the DR of the image is way beyond what any of these cameras can cope with. So low iso DR covers a significant minority of important pictures. Here's an example: http://www.cyberphotographer.com/s/20050122pitnzit/42.jpg
Pictures like that can certainly benefit from greater ability to hold higlights. Black and white film was amazingly tolerant of overexposure, and if you wanted to 'burn in' blown highlights, you usually could.
Here's another example of the kind of image where ability to rescue deep shadows can be very helpful in stopping highlights from blowing out. My 1Ds3 did a superb job here, and allowed me to retain the bright sky texture while seeing deep into the shadows and distinguishing clean blacks. The 5D2 will not do this nearly as well: http://www.cyberphotographer.com/m/20080726tommatthewstessawatson/bw/0014.jpg ...Show more →
Is there any reason why you didn't use any fill flash in condition such as those two samples that you show us?
Because I don't like how it looks. In fact the first shot does appear to have some fill-flash, but I almost always prefer the look of natural light. It's nice to get some atmosphere, so if on camera flash can be avoided, I do. There are countless applications in which photographers want plenty of dynamic range. It's no good trying to pretend that the need for DR doesn't exist or can be obviated by Photography 101 techniques. DR matters.
yauyi wrote:
When done properly I think fill flash can work wonder and look very natural.
I obviously do it very badly then...
Seriously - I would love to see an example of it looking as good as ambient light - I've only been watching 25 years. Different photographers like different things. This is a complete distraction from the topic which is about the 5D2's shadow noise and compromised DR. Fill flash is not a general solution to that problem, no matter how crap you don't think it looks.
p.14 #10 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
This is a complete distraction from the topic which is about the 5D2's shadow noise and compromised DR.
Talking about distractions... Are we done yet or are more lenscap/PS level samples required? I mean, do we have enough data that points towards the criss cross pattern being a normal thing for the 5D2
Unfortunately I couldn't get hold of John and Will's files...
p.14 #11 · Are you seeing 5D2 Criss-Cross Pattern?
It would be nice to see some more samples, but so far there have been no results which contradict the performance I'm seeing from mine. Add that to the poll, which shows that the majority are seeing the pattern with various convertors, and I'm satisfied that this is what a factory spec 5D2 does.
I would still like to see how an original 5D does in the lens cap test.
p.14 #12 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac, sure your planned purchase will serve you well, and i agree that DR matters, in some real wolrld situations 11 stops over 10 makes the difference.
However, I honestly believe that Canon were sincere in their marketing and they didn't see this issue coming.
I know for sure that DXOmark has measured the DR and SNR of the 5DMKII and 1DsMKIII practically identical at ISO 100 and 200, and I also know for sure that their measurement method would pick up this pattern noise as decreased DR and SNR.
But what does that help, when nobody else get such a camera? I belive that there are a few ones out, and time will show. Maybe this thread will turn, maybe not.
p.14 #13 · Are you seeing 5D2 Criss-Cross Pattern?
alundeb wrote:
I know for sure that DXOmark has measured the DR and SNR of the 5DMKII and 1DsMKIII practically identical at ISO 100 and 200, and I also know for sure that their measurement method would pick up this pattern noise as decreased DR and SNR.
You have a lot more confidence in DXOmark's methods than I do. I raised the question of DXOmark's quantification a while back. Their noise measurements contain no assessment of the SCALE and ENTROPY of noise, and those are very important factors, because the human eye is a pattern recognition supercomputer. We really notice patterns.
p.14 #15 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
You have a lot more confidence in DXOmark's methods than I do. I raised the question of DXOmark's quantification a while back. Their noise measurements contain no assessment of the SCALE and ENTROPY of noise, and those are very important factors.
True, that way their numbers say nothing about how intrusive the noise looks. They use standard deviation of the histogram, and the standard deviations of the files compared here are surely not the same. There is about 1 stop difference.
p.14 #16 · Are you seeing 5D2 Criss-Cross Pattern?
yauyi wrote:
is it not an option to return your 5D2 for an exchange? or send it in to Canon for service or replacement?
I have no evidence that it isn't working according to spec. There is some evidence on this thread that even if you send it back to Canon and they claim to fix it, it still produces results exactly like my camera. I also love...
iso 6400 - 25600
size/weight
1080p
3 C-modes
face-detection
the monitor
sraw1
and so on. It will work very well for me, except in the minority of pictures when I am struggling to retrieve shadow data, and 50 iso looks like it might help me there. Until I have evidence that my camera isn't how it should be, what's the point in sending it back?
p.14 #17 · Are you seeing 5D2 Criss-Cross Pattern?
alundeb wrote:
True, that way their numbers say nothing about how intrusive the noise looks.
Exactly. Imagine an almost noiseless camera which had a really really noisy 100 pixel wide McDonalds logo in the middle of the frame. It would score very high in DXOmark's graphs, but be a useless camera. Don't believe what you read. Unless I wrote it.
p.14 #19 · Are you seeing 5D2 Criss-Cross Pattern?
I'll watch out
Another thing DXO were very careful about in their method, was that their background was clean, using glass rather than a cardboard.
But there has been some odd results from that side, so one must be careful.
And one back:
If you remember the first unsharp test of the new 24mmII on the-digital-picture.com, I was one of the first ones out saying I didn't trust that test. (maybe that was in a DPR forum) At last, they found the problem was DPP using NR.
Sure you don't have criss cross pattern running inside your brain... iac? 
